Answers are grounded in this paper and cite their sources.
A robust principal component analysis examining two-years of monitored data from within 143 UK schools as part of the “Schools’ Air quality Monitoring for Health and Education” dataset sheds new light on the temporal and geographical trends that influence the schools’ air quality. Minutely concentration measurements of two indoor air quality markers, particulate matter and carbon dioxide, were analysed. Significant inter-school variations were evident in both species but the primary source of variations within the particulate matter concentrations were temporal trends. Analysis links these temporal trends to variations in outdoor particulate matter concentrations — highlighting the importance of outdoor air quality for the air quality in UK schools. Moreover, clear geographical signatures are identified within the particulate matter concentrations in schools, broadly increasing both South–North and East–West; with more subtle but intelligible trends overlaid within each. Analysis of the carbon dioxide data identifies variations on a range of time scales from seasonal trends, associated with the changing levels of ventilation as UK schools open their windows less in cooler wintertime conditions, down to inter-day variations as carbon dioxide levels increase during the school morning. This study provides the first evidence as to the insights that can be gained when large-scale longitudinal monitoring of indoor spaces, in this case schools, is carried out. The insights that were determined from analysis of two-years of data from the SAMHE schools could not have been realised by analysis of smaller datasets. • CO 2 and PM 2.5 data measured in 143 UK schools over academic years 2023–2025 is analysed • RPCA reveals the sources of school-wide and temporal variation within the data • For PM 2.5 data, temporal variations were the largest, driven by changes in outdoor PM 2.5 • But, for CO 2 , a pollutant generated indoors, school-wide variation was the largest • These methods reveal patterns that are not obvious from analysis of the raw dataset.
No takes yet. Share an insight, caveat, or question.
Handy et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: